Regression and Learning with Pixel-wise Attention for Retinal Fundus Glaucoma Segmentation and Detection

TL;DR

This paper introduces two deep learning algorithms utilizing pixel-wise attention mechanisms for accurate glaucoma detection and optic disc/cup segmentation from retinal fundus images, outperforming existing methods.

Contribution

It presents novel attention-based convolutional neural networks with specific strategies for pixel-wise feature learning in glaucoma diagnosis.

Findings

Achieved state-of-the-art performance on validation dataset.

Demonstrated effectiveness of attention mechanisms in medical image segmentation.

Provided open-source code for reproducibility.

Abstract

Observing retinal fundus images by an ophthalmologist is a major diagnosis approach for glaucoma. However, it is still difficult to distinguish the features of the lesion solely through manual observations, especially, in glaucoma early phase. In this paper, we present two deep learning-based automated algorithms for glaucoma detection and optic disc and cup segmentation. We utilize the attention mechanism to learn pixel-wise features for accurate prediction. In particular, we present two convolutional neural networks that can focus on learning various pixel-wise level features. In addition, we develop several attention strategies to guide the networks to learn the important features that have a major impact on prediction accuracy. We evaluate our methods on the validation dataset and The proposed both tasks' solutions can achieve impressive results and outperform current state-of-the-art methods. \textit{The code is available at \url{https://github.com/cswin/RLPA}}.